Fivetran supports PostgreSQL data warehouse as a destination. Postgres is not typically used as a data warehouse because it is a row-based database more suited to transactional use-cases. Large analytical querying is much faster on columnar warehouses. Postgres can, however, be effective as a warehouse for smaller data volumes.
Fivetran supports connecting with different PostgreSQL implementations. We have specific instructions for connecting to each:
Type transformation mapping
The data types in your PostgreSQL warehouse follow Fivetran's standard data type storage.
We use the following data type conversions:
|Fivetran Data Type||Destination Data Type||Notes|
|INSTANT||TIMESTAMP WITH TIME ZONE|
|LOCALDATETIME||TIMESTAMP WITHOUT TIME ZONE|
|STRING||VARCHAR or TEXT||VARCHAR if
Note: If both your data source and destination are based on PostgreSQL, you may observe unexpected replication results because the standard warehouse types do not cover all available types in PostgreSQL. For example, GEOMETRY and JSONB types are not going to be replicated as the same types, because the type transformations that we perform are intended to work for all destinations. Therefore, we convert JSONB into JSON and GEOMETRY into separate x and y columns of DOUBLE.
Data load costs
Whether you are self-hosting PostgreSQL or using a managed service, you will not be charged extra when we load data into your warehouse.
We need the
default_transaction_read_only paramater value to be set at 0 (off) to be able to create schema and tables in your destination. See PostgreSQL documentation for more information.